Ag

Ra

b

a

ARRAA

KBWAGGF

1

ipbmahirotb

sm

0h

Catalysis Today 223 (2014) 3– 10

Contents lists available at ScienceDirect

Catalysis Today

jou rn al hom epage: www.elsev ier .com/ locate /ca t tod

cid catalysed alcoholysis of wheat straw: Towards secondeneration furan-derivatives

.J.H. Grisela,∗, J.C. van der Waalb, E. de Jongb, W.J.J. Huijgena

Energy Research Centre of the Netherlands (ECN), Biomass & Energy Efficiency, P.O. Box 1, 1755 ZG Petten, The NetherlandsAvantium Chemicals B.V., Zekeringstraat 29, 1014 BV Amsterdam, The Netherlands

r t i c l e i n f o

rticle history:eceived 9 January 2013eceived in revised form 30 May 2013ccepted 1 July 2013vailable online 19 August 2013

eywords:iomassheat straw

lcoholysislucopyranosidelucosideurans

a b s t r a c t

The acid-catalysed alcoholysis of wheat straw has been studied in 95% methanol and 94% ethanol (w/w)in the presence of various amounts of H2SO4 and compared to the alcoholysis of wheat straw-derivedorganosolv pulp and commercially available celluloses. Substrate liquefaction and the product distribu-tion were found to depend mainly on the temperature and the amount of H2SO4 added compared to theacid neutralisation capacity (ANC) of the substrate. The process was optimised for the one-step conver-sion of wheat straw into methyl glucosides, defined as the sum of � and � anomers. The maximum totalmethyl glucosides yield from wheat straw was 56 mol-% based on initial glucan after 120 min methanoly-sis at 175 ◦C and 40 mM H2SO4. Concurrently, furfural was formed at 40 mol-% yield based on initial xylan.The solid residue consisted of mainly acid insoluble (pseudo)lignin, humins and minerals. Switching toethanol resulted in a shift from glycosides to furfural, 5-(alkoxymethyl)-2-furfural and levulinates. Addi-tion of MgCl2, as well as substituting H2SO4 by HCl led to poorer biomass liquefaction and lower glucosidesyield presumably due to consumption of protons under the typical reaction conditions. Alcoholysis of

delignified, cellulose-enriched pulp obtained via organosolv fractionation resulted in higher glucosidesyields and more concentrated product streams, as higher glucan loadings are possible and undesiredside-reactions are minimised. Furthermore, organosolv fractionation prior to alcoholysis allows for theseparation and valorisation of the lignin fraction. The glucosides can be separated, e.g. by means of chro-matography, and may be converted into furan building blocks, for example for the production of plasticprecursors, such as 2,5-furandicarboxylic acid.

. Introduction

A growing world population and its persistent drive for increas-ng the standard of living and the growing needs for mobility areutting a large claim on the natural resources. It is commonlyelieved that resources such as clean water, food, energy and rawaterials will become scarce in the foreseeable future. Worries

bout running out of natural resources have been from all ages, butas never led to insurmountable problems due to the adaptabil-

ty and ingenuity of men. These days, declining known oil and gaseserves are a major topic in international discussions. In the lightf anthropogenic global warming an obvious switch to alterna-ive fossil carbon sources for energy production and petrochemicaluilding blocks is not recommendable.

In many instances, a petrochemical building block can beubstituted by natural alternatives. In this light, the U.S. Depart-ent of Energy published two reports identifying promising

∗ Corresponding author. Tel.: +31 88 515 42 16.E-mail address: grisel@ecn.nl (R.J.H. Grisel).

920-5861/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.cattod.2013.07.008

© 2013 Elsevier B.V. All rights reserved.

renewable building blocks that can be produced from biomasssugars [1] and lignin [2]. Furan derivatives are considered to behigh-potential building blocks for bio-based fuels and polymers [1].A key intermediate in the production of biomass-derived furans isHMF (5-(hydroxymethyl)-2-furaldehyde) [3], which can be usedto produce, for example, 2,5-dimethylfuran (DMF) [4] and 2,5-furandicarboxylic acid (FDCA) [5]. DMF is regarded as a potentialbio-gasoline [6] and FDCA can replace the petrochemical tereph-thalic acid, a precursor to the polyester PET, to make clothing andplastic bottles [7].

HMF is an intermediate in the conversion of hexoses to lev-ulinic acid and is produced in aqueous solution via acid-catalyseddehydration of hexoses, such as fructose and glucose [8–12].Herein, glucose first needs to be converted to fructose by iso-merisation. Under typical reaction conditions, the glucose–fructoseisomerisation is rate-limiting, and HMF yields from fructose arehigher than starting from glucose [11–13]. Addition of a base

is beneficial for the production of HMF from glucose, by selec-tively accelerating the isomerisation reaction [8,14]. Increasingthe temperature will speed up both glucose isomerisation anddehydration to form HMF, but even more the successive HMF

4 lysis T

pa

saipaasfa[iaos

(pcHuhphhicaya1iy[aHetitos

cTbhictgfiafptovvnaa

p

R.J.H. Grisel et al. / Cata

olymerisation and decomposition yielding levulinic and formiccid [9].

Preferably, HMF is produced from non-edible natural resources,uch as cellulose. When starting from lignocellulosic biomass, suchs wheat straw or wood chips, or from cellulose the first steps depolymerisation. Typical aqueous acid hydrolysis methods forroducing HMF from (ligno)cellulose are based on high temper-tures. Consequently, the rate of HMF degradation via hydrolysisnd polymerisation reactions is high and the yield is limited. Inhort, HMF is not stable under the typical acidic conditions requiredor its formation from cellulose or lignocellulosic biomass in waternd the maximum yield from d-glucose is typically below 30%8,11,15]. The yield can be improved by reactive extraction of HMFn a non-aqueous phase, preventing HMF degradation by hydrationnd subsequent decomposition. Fan et al. [13] reported a 68% yieldf HMF from glucose in aqueous methyl isobutyl ketone, using theolid heteropoly acid salt Ag3PW12O40.

Recently, organic solvents such as N,N-dimethylacetamideDMA) [11,13,16] and ionic liquids [11,17–19] were shown to haveotential for the selective production of HMF from glucose andellulose, especially in the presence of a metal chloride catalyst.erein, the improved yields originate from improved cellulose sol-bility, enhanced glucose–fructose isomerisation, repressed HMFydration, or a combination of those. The influence of loosely ion-ared halide ions, such as chloride, was explained by effectiveydrogen bond disruption within the solid thus enabling celluloseydrolysis at reduced temperatures, while simultaneously lower-

ng HMF degradation. In DMA/LiCl, the maximum HMF yield fromellulose was circa 30% in the presence of Zr(O)Cl2 or CrCl3 [16]. For

combination of CrCl3, HCl and LiBr in DMA/LiCl a maximum HMField of 37% was reported [11]. In both cases, the presence of largemounts of 1-ethyl-3-methylimidazolium chloride ([EMIm]Cl] and-butyl-3-methylimidazolium ([BMIm]Cl) resulted in a HMF yield

ncrease to 54–57%. A mixture of CuCl2 and CrCl2 in [EMIm]Clielded a maximum HMF of 55% from cellulose after 8 h at 120 ◦C19]. After 1 h at 140 ◦C, maximum HMF yields of respectively 53%nd 42% were reported for cellulose and corn stover in CrCl3 andCl containing [EMIm]Cl [11]. While in pure ionic liquids HMF canasily be produced from fructose using a variety of metal chlorides,he production from glucose and polymers thereof seems to specif-cally require the presence of chromium chloride [17,18]. Next tohe environmental disadvantages of chromium salts, the high costf ionic liquids limits the industrial application of such solventsystems.

Alternatively, in acidified methanol and ethanol hexose-ontaining material can be converted into stable HMF ethers [20].hese ethers can be isolated and converted further into valuableuilding blocks, such as FDCA [7,12]. Thus far, etherification of HMFas been shown successful with good yield and selectivity start-

ng from fructose [21]. Etherification starting from glucose remainshallenging. Within the national Dutch Catfur project, ECN, Avan-ium Technologies and Utrecht University cooperate to developlucose-based, second generation furan-derivatives. Herein, therst step is liquefaction of lignocellulosic biomass using short-chainlcohols. Garves reported that the main products from the celluloseraction are alkyl glucosides or alkyl levulinates, depending on therocess conditions [22]. Under water-lean conditions the forma-ion of levulinic acid and humins is reduced [23]. The suppressionf humins formation is a major benefit in the economics of biomassalorisation, since humins are typically considered suitable for low-alue applications only, such as heat and power generation. It isoteworthy that obtained alkylated sugars are highly soluble in

lcohol and may be further converted to yield the same HMF etherss from glucose and fructose.

Liquefaction of lignocellulosic biomass in alcohols results inroduct streams that also contain solubilised lignin fractions and

oday 223 (2014) 3– 10

hemicellulose derivatives. Their presence may lead to a loss of sug-ars due to undesired condensation reactions. Alternatively, priorto alcoholysis the biomass can be fractionated. Organosolv frac-tionation enables separation of the three main constituents oflignocellulosic biomass, viz. lignin, hemicellulose and cellulose[24,25]. The pulp can be used for the production of alkyl glu-cosides, whereas the lignin and hemicelluloses can be valorisedseparately. When performing alcoholysis on delignified pulp, unde-sired side-reactions are expected to be minimised, the sugar yieldsincreased and more concentrated product streams will be possiblecompared to alcoholysis on the whole biomass. Whereas methanolseems the solvent of choice in alcoholysis of hexose-containingmaterial, ethanol is interesting when starting with lignocellulosicbiomass, due to the higher lignin solubility [26]. Besides, ethanolis considered less toxic and can easily be obtained from renewableresources. Moreover, in acid catalysed alcoholysis at elevated tem-perature the use of ethanol is preferred over methanol because ofits lower tendency to form dialkylether [27].

Previously, H2SO4, HCl, and MgCl2 have been compared andtested as catalyst in organosolv fractionation of wheat straw inethanol/water (60% ethanol, w/w) [25]. Herein, the effect of theacid catalysts was found to be primarily due to their effect on thepH of the organosolv liquor rather than the type of anion. H2SO4 wasfound to be a weaker acid than HCl in 60% (w/w) aqueous ethanol.The use of MgCl2 was found to enhance fractionation and enzy-matic digestibility of the resulting pulp, however, to a lesser extentthan both acid catalysts.

In this study cellulose containing material is processed in acid-ified methanol and ethanol, targeting maximum total glucosidesyields. Herein, no difference is made between the yields of the �and � anomer of the glucosides; both anomers can be used simul-taneously as precursor for the formation of furan-derivatives. Theeffect of substrate, catalyst concentration, temperature and reac-tion time are studied, as well as the effect of HCl and MgCl2 addition.

2. Experimental

2.1. Materials

Wheat straw was selected as a representative lignocellulosicbiomass substrate, because it is an abundant agricultural residue inEurope. The wheat straw used in this study was dried winter wheatstraw grown in the vicinity of Delftzijl, the Netherlands, and cut to<2 cm pieces. The average moisture content was ∼8% (w/w) drybiomass. The bulk composition of the wheat straw was: 8.4% waterextractives, 2.0% ethanol extractives, 36.9% glucan, 19.9% xylan,1.9% arabinan, 0.7% galactan, 0.2% mannan, 0.1% rhamnan, 16.7%acid-insoluble lignin, and 1.1% acid-soluble lignin. The ash contentwas 6.1%. The elemental composition of the biomass was 44.3% car-bon, 42,7% oxygen, 5.4% hydrogen, 2.2% silicium, 0.2% potassium,0.2% calcium (other elements <0.1%) [25]. All percentages are basedon dry biomass weight. The straw was stored dry in closed bags atroom temperature until use. The material was used without furthertreatment.

Cellulose-enriched lignocellulosic pulp (68.9%, w/w glucan;7.9%, w/w xylan, 15.7%, w/w acid-insoluble lignin, and 0.4%, w/wacid-soluble lignin) was obtained from organosolv pulping of wheatstraw [28]. For this, the wheat straw was mixed with 50% (w/w)aqueous ethanol in a ratio of 10 L/kg dry biomass and the slurrywas heated to 210 ◦C while being mixed. The reactor was keptisothermal for 90 min and then cooled to below 40 ◦C. After organo-

solv treatment, the product was filtered. The solid fraction waswashed with 50% (w/w) aqueous ethanol and dried in vacuo at50 ◦C. The pulp was stored in closed vessels until further use.As lignin and hemicellulose-free reference samples Avicel PH-101

lysis T

m(S

wawstt�ec

2

2

rcccdmsa(bwssTc

asHmaatroCwd4Bt

2t

rmlawtt�

rfAt

R.J.H. Grisel et al. / Cata

icrocrystalline cellulose (∼50 �m particle size) and �-cellulose∼50–500 �m particle size distribution) were purchased fromigma–Aldrich and used without further purification.

Methanol (99.9%, w/w), ethanol (92.6–95.2%, w/w, average 94%,/w used in the calculations), H2SO4 (98%, p.a.), HCl (32%, p.a.)

nd MgCl2·6H2O (p.a.) were all purchased from Merck and usedithout further purification. Alkyl-glycosides used in the analy-

is standards were obtained from Carbosynth, Berkshire, UK, withhe exception of �-ethyl-d-xylopyranoside. To be able to determinehe elution time of �-ethyl-d-xylopyranoside, a mixture of � and-ethyl-xylosides was prepared by heating xylose in 94% (w/w)thanol containing 50 mM H2SO4 to 70 ◦C and injected onto theorresponding column.

.2. Alcoholysis process

.2.1. Reactor set-upAlcoholysis experiments were performed in two different

eactor types. Screening experiments on the effect of H2SO4oncentration and the alcoholysis experiments with alternativeatalysts (HCl, MgCl2) or alternative substrates (organosolv pulp,ellulose) were carried out in parallel 125 mL batch reactors (acidigestion bomb type 4748, SS 316 with Teflon liner, Parr Instru-ent Company, Moline, IL). The substrate was suspended in the

olvent and subsequently the catalyst or a mixture of catalysts wasdded. The closed reactor vessels were placed in a heating blockadapted RS600, Thermo Fisher Scientific, Rochford, UK) and thelock temperature was set to the desired temperature. The slurriesere stirred by a magnetic bar, located at the bottom of each ves-

el (1000 rpm). After 180 min the power to the heating blocks waswitched off and the reactors were cooled while kept in the blocks.ime at temperature within 5 ◦C of the set-point was estimated atirca 30 min [25]. This reactor type is referred to as reactor A.

Screening experiments on the combined effect of temperaturend acid concentration, as well as the methyl-glucosides optimi-ation experiments were carried out in an autoclave reactor (0.5 Lastelloy Kiloclave, Büchi Glas Uster AG, CH). The substrate wasixed with either 95% (w/w) methanol or 94% (w/w) ethanol, bal-

nce H2O. Sulphuric acid was added in catalytic amounts. Undergitation (100 rpm, propeller stirrer), the slurry was heated to reac-ion temperature and kept at this temperature for 30–150 min. Theeactor was equipped with a standard sampling system consistingf a Hastelloy dip tube with a 7 micron filter (Büchi Glas Uster AG,H) to allow sampling at temperature and pressure. The samplesere quenched in an ice-water cooled coil attached directly to theip tube. After reaction, the reactor was cooled actively to below0 ◦C in less than 30 min. This reactor type is referred to as reactor. Due to divergent heating and cooling transients of both reactors,he results can only be compared within their series.

.2.2. Screening experiments on the effect of acid dose,emperature, and incubation time

The screening experiments on the effect of acid dose were car-ied out in reactor A with 6 g wheat straw (as received) in 45.6 gethanol and 2.4 g H2O obtaining 95% (w/w) methanol and a

iquid-to-solid ratio of 11.3 L/kg (dry weight). Subsequently variousmounts of H2SO4 were added (0–50 mM). The block temperatureas set at 200 ◦C. The effect of the acid concentration was studied in

erms of solvolysis performance and total glucosides yield. Herein,he glucosides yield is the sum of �-methyl-d-glucopyranoside and-methyl-d-glucopyranoside anomers being produced.

The methanolysis yielding the highest glucosides yield was

epeated at 25 g wheat straw scale (as received) in reactor B 200 ◦Cor 30 min. The solvent to substrate ratio was 11 L/kg (dry weight).dditional experiments were carried out in which the reaction

emperature was varied between 160 and 200 ◦C. The incubation

oday 223 (2014) 3– 10 5

time and acid concentration were varied as well, thus taking intoaccount combined effects. The reaction conditions are summarisedin Table 1.

After cooling, the product was centrifuged and the liquidwas analysed with a combination of HPLC, HPAEC and GC–MStechniques as described in Section 2.3. The solid residue waswashed, dried and weighed to calculate the pulp yield or lique-faction efficiency. The resulting pulps of the top-three performingexperiments in terms of glucosides yield were analysed for theircomposition and the amount of non-converted glucan was deter-mined according to the methods described in Section 2.3. Basedon the combined results of glucosides yield, pulp yield and pulpcomposition, a set of process conditions was defined and taken asthe starting point for further optimisation to yield a maximum ofmethyl glucosides from wheat straw.

2.2.3. Optimising glucosides yieldBased on the screening experiments a set of condition was

defined in terms of processing temperature, incubation time andadded acid dose. A set of optimisation experiments was carriedout in reactor B with 30 g wheat straw (as received) in 95% (w/w)methanol (11 L/kg dry weight) at fixed temperature, ranging from175 to 190 ◦C, and a H2SO4 concentration, ranging from 25 to40 mM. After 105 min at temperature the sampling line was flushedwith 5 mL and consequently a 5 mL sample was drained from thereactor and quenched in ice water to stop the reaction and condensethe product mixture. This was repeated at a 15 min interval until4 samples were isolated per set of conditions. The liquid sampleswere analysed for methyl glucosides by HPAEC-PAD, as describedin Section 2.3. The set of conditions yielding the highest amount ofmethyl glucosides in the sampling experiments were verified singlebatch experiment, to eliminate possible effects of volume changesand a different cooling transient as a result of the sampling in theoptimisation experiments.

2.2.4. Effect of ethanol and chlorideThe usability of ethanol for the production of glucosides from

wheat straw was evaluated qualitatively by comparing the resultsof ethanol alcoholysis at 200 ◦C and 30 mM H2SO4 with thoseobtained after methanolysis under similar conditions. The effect ofchloride on the alcoholysis of wheat straw was tested by the addi-tion of 50, 100 and 200 mM MgCl2 and, in a separate series replacing30 mM H2SO4 by 60 mM HCl. The MgCl2 experiments were carriedout in 95% (w/w) methanol at 180 ◦C in the absence and presence ofH2SO4. The HCl experiments were performed in 94% (w/w) ethanolat 200 ◦C. All results were compared to a H2SO4 reference.

2.3. Analyses

2.3.1. Summative composition of wheat straw and obtained solidsThe ash content of the wheat straw was determined by cal-

cination according to NREL/TP-510-42622 [29]. The elementalcomposition was measured with an elemental analyser (Carlo ErbaInstruments FLASHEA1112, Wigan, UK) (C, H, N, and O), ion chro-matography after bomb combustion (Cl), and inductively coupledplasma atomic emission spectroscopy (ICP-AES) (other elements).The further composition of the raw materials and obtained solidresidues, i.e. lignin, hemicellulose and cellulose, were analysedusing a modified hydrolysis protocol based on TAPPI methods T 222and 249 [30]. In this, the solids were hydrolysed in subsequently12 M H2SO4 at 30 ◦C for 1 h and 1.2 M at 100 ◦C for 3 h. The amountof solid residue, i.e. acid-insoluble lignin and ash, was determined

gravimetrically and the ash content herein was determined by cal-cination. The hydrolysate was analysed for acid-soluble lignin byUV–vis absorption spectroscopy at 205 nm and monomeric reduc-ing sugars by HPAEC-PAD (see Section 2.3.2). In case of the wheat

6 R.J.H. Grisel et al. / Catalysis Today 223 (2014) 3– 10

Table 1Glucan-based product yields after methanolysis of wheat straw.

Entry Temperature (◦C) Incubation time (min) H2SO4 (mM) Solid residue (wt-%) Unconverted substrate and main product groups

Solid glucan (%)b Glucosides (mol-%)c Furans (mol-%) Levulinates (mol-%)

1 200 30a – 84.2 96.8 0.0 0.0 0.02 200 30a 10 72.8 96.0 0.2 0.1 0.03 200 30a 20 33.7 61.7 23.3 0.7 0.74 200 30a 30 14.9 6.0 37.7 7.1 15.35 200 30a 40 15.5 1.6 –d 3.0 26.46 200 30a 50 20.1 0.1 –d 2.5 40.4

7 200 30 30 15.1 6.0 41.0 6.2 18.68 200 60 30 18.0 –d 29.2 3.0 26.49 180 60 30 29.5 –d 35.2 –d –d

10 180 120 30 22.9 28.4 43.9 2.9 7.011 170 120 50 20.3 15.9 44.4 3.3 12.612 160 120 50 32.6 –d 30.6 –d –d

13 160 120 150 22.4 –d 33.2 –d –d

a Performed in reactor A: estimated time at temperature...

se4stm

2

cdaaHAvcwf(5nei

eoeffw

(faw(t0

(W(d

b Unconverted glucan as percentage of glucan present in the wheat straw samplec �-Methyl-d-glucopyranoside/�-methyl-d-glucopyranoside molar ratio 1.7–1.8d Not analysed.

traw, the extractives were removed with two successive Soxhletxtractions using water and ethanol according to NREL/TP-510-2619 [29] prior to hydrolysis. The moisture content of the solidamples was measured with a halogen moisture analyser (Met-ler Toledo HR83, Columbus, OH). Unless otherwise indicated, all

entioned solid masses are on dry weight basis.

.3.2. Composition of alcoholysis liquorsThe alcoholysis liquors were analysed for their acidity and

ontent of glycosides, monomeric reducing sugars, sugar dehy-ration products, and organic acids. Analysis of the alphand beta anomers of the glucosides (alkyl-d-glucopyranosides)nd xylosides (alkyl-d-xylopyranoside) was performed withigh Performance Anion Exchange Chromatography with Pulsedmperometric Detection (HPAEC-PAD, ICS3000, Dionex, Sunny-ale, CA) equipped with a CarboPac MA1 column and a postolumn addition of 0.3 mL/min 0.25 M NaOH. A gradient of NaOHas used as eluent (0.25 mL/min): 50 mM (0–5 min), increasing

rom 50 to 425 mM (5–37.5 min), increasing from 425 to 800 mM37.5–40 min), 800 mM (40–60 min), and decreasing from 800 to0 mM (60–62 min), 50 mM (62–70 min). With this set-up it wasot possible to completely separate ethyl-�-glucopyranoside andthyl-�-xylopyranoside. Due to the low yields of xylopyranosidesn ethanol alcoholysis, the error made herein is small.

Analysis of monomeric sugars was performed by HPAEC-PADquipped with a CarboPac PA1 column and a post column additionf 0.2 mL/min 0.25 M NaOH. A gradient of NaOH was used as elu-nt (0.25 mL/min): 15 mM (0–2 min), 0 mM (2–19 min), increasingrom 0 to 200 mM (19–35 min), 200 mM (35–50 min), decreasingrom 200 to 15 mM (50–51 min) and 15 mM (51–59 min). Lactoseas used as an internal standard.

The concentration of furfural (furan-2-carbaldehyde), HMF5-(hydroxymethyl)-2-furaldehyde), MMF (5-(methoxymethyl)-2-uraldehyde) and EMF (5-(ethoxymethyl)-2-furaldehyde), as wells acetic acid, formic acid, and levulinic acid (4-oxopentanoic acid),as determined with High Performance Liquid Chromatography

HPLC, Agilent 1100 series) equipped with an RI analyser, UV detec-or and a Bio-Rad Aminex HPX-87H column (65 ◦C, 5 mM H2SO4,.60 mL/min).

Analysis of levulinate esters was performed with GC–MS

Trace GC ultra DSQ II, Thermo Scientific, with Phenomenex ZB-

AXplus column) starting at 40 ◦C (0–5 min), increasing to 245 ◦C10 ◦C/min) and 25 min dwell time at 245 ◦C. The samples wereiluted in isopropanol before analysis.

2.3.3. Acid-neutralisation capacitySeveral typical minerals in biomass (fractions) have an acid-

neutralising effect, influencing the effect of added acid catalyst. Theacid-neutralisation capacity (ANC) of the wheat straw was deter-mined using pH dependence leaching tests at pH 2 and pH 4 [31].A mixture of 50 g wheat straw (as received) in 500 g water wastitrated with concentrated HNO3 (65%) until the pH of the solutionreached the desired value. The closed vessel was placed on a rollerbank. After 4, 24 and 48 h of equilibration, a measured amount ofadditional HNO3 was added to retain the desired pH. Hereafter, afurther change in pH is negligible and the total HNO3 consumptionwas determined. The ANC of the wheat straw was 0.10 and 0.26 molH+/kg wheat straw (dry weight) at pH 4 and pH 2, respectively.

3. Results and discussion

3.1. Screening and optimisation of glucosides yield

Initial screening experiments were carried out in reactor A onuntreated wheat straw in 95% methanol (w/w) at 200 ◦C and vary-ing amounts of H2SO4. Table 1 (entry 1–6) shows the influence ofthe H2SO4 concentration on the products yields obtained after alco-holysis of wheat straw in methanol. The solid residue is based on theweight change of dried solids before and after reaction; the productyields are given in mol-% based on initial glucan input. The furansyield is the sum of HMF and MMF yields and the levulinates yieldis the sum of levulinic acid and methyl levulinate yields. Typically,the methylated compounds constitute >90% for both main productgroups.

Based on the degree of solubilisation and glucosides yield, theoptimum concentration of H2SO4 under given reaction conditionswas 30 mM. Lower acid concentrations led to poor glucan con-version and a limited glucosides (and furans) yield. Higher acidconcentrations as well as doubling the incubation time (Table 1,entry 8) led to high yields of levulinates as well as an increasein solid residue after methanolysis, implying the formation ofinsoluble humins. Moreover, at the higher acid concentrationsthe pressure build-up in the reactor was considerably higherthan may be expected from the vapour pressure equilibrium ofwater/methanol mixtures. This was found to be related to the for-mation of the more volatile dimethyl ether from the acid-catalysed

dehydration of methanol. In the range studied, the formation ofdimethyl ether was kinetically limited and methanol losses dueto the formation of dimethyl ether were found to vary roughlyfrom 5 to 30%, depending on the reaction conditions. The glucose

R.J.H. Grisel et al. / Catalysis Today 223 (2014) 3– 10 7

fter 12

c≤

sf5≥

iTs1iHtdaee(

pT2wwstsa

rwwtaliBpgiw

m1Ha

Fig. 1. Methyl glucosides yield a

oncentration in the liquors was <0.3% at H2SO4 concentrations20 mM and around 4% at higher acid doses.

At a H2SO4 concentration of 30 mM the xylan fraction of theubstrate was completely solubilised. The major product was fur-ural (>43 mol-% based on xylan). The xylose yield dropped from.9% in the absence of H2SO4 to below 1% at a H2SO4 concentration20 mM.

The pH of the resulting liquors was around 2 and lower for exper-ments carried out with an initial H2SO4 concentration ≥20 mM.he ANC of the wheat straw was 0.26 mol H+/kg at pH 2, corre-ponding to an effective H2SO4 concentration reduction of about2 mM in these experiments. Based thereon, the sudden increase

n acid-catalysed glucan solvolysis between 10 and 20 mM initial2SO4 concentration is understandable, since at these concentra-

ions not all added acid will have been neutralised. It is noted thaturing solvolysis of biomass some organic acids are being produceds well, such as acetic acid, formic acid and levulinic acid. How-ver, in alcohol-rich environments large part of the organic acids issterified and will not contribute to the acidity of the final mixtureesterification typically >90%).

The screening experiments were extended by lowering the tem-erature and varying the incubation time and acid concentration.he highest methyl glucoside yields were obtained after 30 min at00 ◦C with 30 mM H2SO4 (Table 1, entry 7), after 120 min at 180 ◦Cith 30 mM H2SO4 (Table 1, entry 10) and after 120 min at 170 ◦Cith 50 mM H2SO4 (Table 1, entry 11). The selectivities for gluco-

ides were 48%, 61%, and 53% respectively. This corresponds withhe trends observed for levulinates yield, indicative for the reactioneverity, and the solid residues. The extent of methanol dehydrationt temperatures below 180 ◦C seemed to be limited.

Compositional analysis of the three corresponding solidesidues showed that nearly all the ash ends up in the residue,hereas no xylan and virtually no acid-soluble lignin (<0.2%, w/w)ere detected in the solids. The amount of acid-insoluble lignin in

he solids was reduced to 32% of the initial amount after alcoholysist 180 and 200 ◦C and 44% after alcoholysis at 170 ◦C, suggesting aower delignification at lower temperature. This fraction will alsonclude lignin-condensates and acid-insoluble humins, if formed.ased on the work by Hu et al. [23], however, the amount of huminsroduced under these conditions is probably limited. The amount oflucan in the solid, expressed as a percentage of the initial amount,s tabulated in Table 1. The mass balance of the composition analysis

as >95% for all solids.Based on results of the screening experiments an experimental

atrix was defined in which the temperature was varied between80 and 190 ◦C and the acid concentration between 25 and 35 mM2SO4. Subsamples were collected after 105, 120, 135 and 150 minnd analysed for methyl glucosides. During the methanolysis

0 min (left) and 150 min (right).

experiments the amount of solvent changes, as a result of hydrol-ysis and condensation reactions, especially when dimethyl ether(DME) is formed. At typical methanolysis conditions, methanol lossdue to the formation of dimethyl ether is estimated to be 5–20%[32]. Net water will be produced and methanol will be consumed,resulting in a lower volume and total weight of the solvent. Sinceduring sampling the amount of remaining liquid cannot be quan-tified, the yields are given in g per kg initial solvent. It should benoted that sampling itself also slightly reduces the total amountof solvent, affecting the glucosides concentration, especially aftermultiple samplings. Consequently, the concentrations found forlonger incubation times will be relatively larger. The concentra-tions were corrected for the solvent loss due to sampling and theresults after 120 and 150 min incubation are depicted in Fig. 1.

The optimum yield per temperature follows the diagonal fromhighest temperature and lowest acid concentration to the lowesttemperature and highest acid concentration. At less severe reactionconditions, i.e. lower temperature and/or lower acid concentrationan increase in incubation time led to increased methyl glucosidesproduction, whereas more severe reaction conditions, i.e. highertemperature and acid concentration, led to a decrease in yieldwith increasing incubation time. The increase is explained by pre-vailing alcoholysis of the remaining glucan in time; the decreaseis a result of enhanced and dominating glucosides degradation,yielding higher amounts of furans and, especially, levulinates. Thetrend implies that an optimum can be found along, or in thevicinity of the mentioned diagonal. Indeed, after 150 min the high-est yield, 25.9 g/kg, was obtained at 185 ◦C with 30 mM H2SO4.The highest yield after 120 min was 26.2 g/kg, obtained at 180 ◦Cwith 35 mM H2SO4. Although the direction towards optimum reac-tion conditions is clear, no overall maximum glucosides yield wasfound in this set of experiments. Therefore, the optimisation setwas extended along the diagonal by performing methanolysisat 175 ◦C with 35 and 40 mM H2SO4 and at 180 ◦C with 40 mMH2SO4.The highest methyl glucosides yield was observed after120 min at 175 ◦C with 40 mM H2SO4 (Table 2). Although an abso-lute maximum in methyl glucosides yield is still not unambiguouslyestablished, based on the screening experiments (Table 1) it wasconcluded that the optimum reaction conditions for this wheatstraw should be near 175 ◦C and 40 mM H2SO4 for a 120 min processand that further optimisation will result in marginal improvements,at best. The maximum yield was estimated at 27.9 g/kg or 55.5 mol-%, assuming no weight changes in the solvent. The concluding pHof the liquor was 1.5.

To validate the results, methanolysis was repeated using thesame reactor set-up at 175 ◦C with 40 mM H2SO4 without inter-mediate sampling using 30 g wheat straw in 238 g methanol 95%(w/w). After 120 min at temperature the reactor was cooled to

8 R.J.H. Grisel et al. / Catalysis Today 223 (2014) 3– 10

Table 2Optimisation of methyl glucosides yield from wheat straw in reactor B.

Temperature (◦C) Incubation time (min) H2SO4(mM) Methyl glucosides (g/kg)a

175 120 35 26.4175 120 40 27.9180 120 40 23.2175 150 35 24.9175 150 40 25.3

bwscgiOHm

vcx

3

9(ffovesieacr

sS

180 150

a �-methyl-d-glucopyranoside/�-methyl-d-glucopyranoside molar ratio 1.8.

elow 40 ◦C. The concluding pH of the liquor was 1.5 and the liquideight loss due to the consumption of methanol, by methanoly-

is and DME formation was less than 3%. The products yields wereorrected for the liquid mass loss and are depicted in Fig. 2. Thelucosides yield was 53.9 mol-% based on initial glucan, which isn agreement with the data obtained in the optimisation series.ther identified, glucan-derived products were glucose (3.7 mol-%),MF (0.2 mol-%), 5-(methoxymethyl)-2-furaldehyde (3.8 mol-%),ethyl levulinate (11.0 mol-%) and levulinic acid (1.0 mol-%).The solid residue after alcoholysis was virtually free of uncon-

erted xylan. The xylan in the wheat straw has preferentially beenonverted into furfural, 40.6 mol-%. Other products were methyl-ylosides (17.9 mol-%) and xylose (1.4 mol-%).

.2. Methanol versus ethanol alcoholysis

Compared to methanol, alcoholysis of wheat straw at 200 ◦C in4% (w/w) ethanol acidified with 30 mM H2SO4 resulted in a loweralkyl) glycosides yield (Fig. 3). However, due to a relatively highurans production, the sum of alkyl glucosides and alkoxymethyl-urfural yield is higher than in methanolysis. Since the formationf alkoxymethylfurfural is a consecutive step in alkyl glucosidesalorisation, e.g. for the production of 2,5-furandicarboxylic acid,thanol seems a suitable solvent to replace methanol in alcoholy-is of wheat straw, if furan-based products are desired. Moreover,n acid catalysed alcoholysis at elevated temperature the use ofthanol is preferred over methanol because of its lower toxicitynd lower tendency to form dialkylether [27]. The liquefaction effi-iencies in methanol and ethanol were 85.1% and 83.2% (w/w),

espectively.

In ethanol, the production of levulinates is not promoted pere and there are no indications of increased humins formation.till, the higher amount of alkoxymethylfurfural and levulinates

Fig. 2. Product spectrum after methanolysis

40 21.8

observed after alcoholysis in ethanol suggest that the reactionconditions in ethanol were slightly more severe. This is in agree-ment with the lower alkyl xylosides and the higher furfural yieldfound for the xylan fraction of the wheat straw (Fig. 3). The differ-ence in severity experienced at similar acid concentration may beexplained by a difference in behaviour of H2SO4 in both alcohols. Itshould be noted that, for both alcohols, the reaction conditions inthese experiments were not optimised for glucosides production.

3.3. Substrate effect (in ethanol)

After alcoholysis of untreated wheat straw, next to the identifiedcompounds, also minerals, lignin fractions and (other) hemicel-luloses derivatives are present in the product liquors. Obviously,the presence of these species will complicate isolation of thedesired products. However, their presence may also influence thefinal product distribution. For example, dissolved lignin fractionsmay react with furan-like species and form soluble and insolu-ble species [25] and minerals may act as co-catalysts or protonscavengers. To visualise the possible effects of reduction of typicalnon-carbohydrate constituents from lignocellulosic biomass, alco-holysis was performed on pure cellulose and cellulose-enrichedpulp (glucan 68.9%, w/w) obtained from organosolv treatment. Theresults are compared with the results obtained with untreatedwheat straw containing 36.9% (w/w) glucan (Fig. 4).

At similar acid concentration the sugars and furans yield forboth the pulp and wheat straw were significantly higher than thoseobtained from Avicel PH-101 and �-cellulose (Fig. 4). Apparently,the (pure) celluloses experience more severe reaction conditions,

in which relatively large amounts of levulinates (and carbonaceousresidue) are formed. This is explained by the absence of acid neu-tralising minerals, resulting in a higher net acid concentration. Thedifferences found in concluding pH of the resulting liquors, pH 1.4

of wheat straw at variable conditions.

R.J.H. Grisel et al. / Catalysis Today 223 (2014) 3– 10 9

Fig. 3. Products yield of wheat straw alcoholysis in

FH

ft

ot(otaopcapeffo

TE

ig. 4. Products yield of different substrates: alcoholysis in ethanol and 30 mM2SO4 (reactor A).

or the celluloses and pH 1.8 for the pulp and wheat straw, supporthis assumption.

Under studied conditions, the total glucosides molar yields fromrganosolv pulp and untreated wheat straw were comparable, buthe furans and levulinates yields from pulp were significantly lowerFig. 4). The relatively lower yields of furans and levulinates, as sec-ndary products, do not correspond with the lower ANC found forhe pulp, 0.19 mol H+/kg versus 0.26 mol H+/kg for the wheat strawt pH 2. At similar substrate loading, however, as a direct resultf the cellulose enrichment, the pulp led to significantly higherroduct concentrations. For example, the total ethyl glucosidesoncentration in the liquor was twice as high. Higher glucosidesnd/or furans concentrations are expected to result in increasedroduct losses due to condensation reactions, especially in the pres-

nce of (residual) lignin. Hereby, furans will be consumed and theormation of levulinates will be reduced, resulting in a lower yieldor both compounds, as is observed for the pulp (Fig. 4). The amountf solid residue obtained after alcoholysis of the pulp, 18.2% (w/w),

able 3ffect of MgCl2 and HCl on the alcoholysis of wheat straw.

T (◦C) Catalyst Solvent (balance H2O) Solid residue (wt-%

180 – 95% methanol 89.8

180 –a 95% methanol 70.4

180 40 mM H2SO4 95% methanol 21.4

180 40 mM H2SO4a 95% methanol 52.7

200 30 mM H2SO4 94% ethanol 15.3

200 60 mM HCl 94% ethanol 52.1

200 100 mM HCl 94% ethanol 48.2

a Added 100 mM MgCl2.b Not analysed.

methanol and ethanol at 200 ◦C (reactor A).

is only slightly higher than that of wheat straw, 16.8% (w/w). There-fore, the larger part of the condensation products, if formed, seemsoluble in the resulting liquor.

3.4. Effect of chloride

The effect of MgCl2 addition is ambivalent. In the absenceof H2SO4 the presence of MgCl2 promoted alcoholysis, whereasin the presence of 40 mM H2SO4 it had a clear inhibiting effect(Table 3). Compared to a non-acidified experiment, addition of100 mM MgCl2 resulted in a decrease in solid residue after reac-tion from 89.8 to 70.4%, as well as a decrease in concluding pHfrom 6.6 to 4.9. Paszner and Cho [33] hypothesise that adsorptionof Mg2+ on carboxylic acid groups of the biomass causes a releaseof protons, thus lowering the pH. It was also suggested that MgCl2effectively hydrolyses the hemicellulose and lignin fraction. The lat-ter hypothesis would explain an increase in solvolysis of the wheatstraw without an increase in soluble glucan derived products. If theeffects are related is not known.

Another explanation may be that this effect is originating fromdissociation of hydrated magnesium cations, releasing protons intothe solution according to: [Mg(H2O)6]2+ = [Mg(H2O)5(OH)]+ + H+.Hydrated magnesium cations are only weakly acidic. The pKain water is about 11.2 at 25 ◦C [34]. Upon addition of H2SO4the measured pH of the mixture rapidly drops below the pointthat deprotonation of Mg-hydrate cation, and hence its promot-ing effect, is no longer expected. Quite the reverse, the presenceof MgCl2 was found detrimental to the alcoholysis in acidifiedmethanol and the methyl glucosides yield. At the same time it wasnoticed that, after reaction, the concluding pH of the liquor was 1.5in the absence and 2.2 in the presence of 100 mM MgCl2. Compared

to the MgCl2-free experiments a lot of protons, initially present,must have been consumed. The consumption of H+ will suppressthe alcoholysis reaction. One such a mechanism consuming H+

may be chlorination of methanol: CH3OH + Cl− + H+ = CH3Cl + H2O.

) Main product groups

Glucosides (mol-%) Furans (mol-%) Levulinates (mol-%)

<0.1 –b –b

0.4 0.2 0.144.6 2.8 10.9

3.4 0.8 0.231.9 19.3 10.3

3.2 1.4 <0.34.7 2.0 <0.3

1 lysis T

A1

Hcyca

4

amtemo

ascsa

totfrtT

ms

A

tVb(na

R

[

[

[

[

[[

[

[

[

[

[[

[

[

[

[

[

[[

[

[

[

acid/base addition, 2005.[32] T.A. Semelsberger, R.L. Borup, Journal of Power Sources 152 (2005)

87–96.

0 R.J.H. Grisel et al. / Cata

t given reaction conditions it made little difference whether 50,00 or 200 mM MgCl2 was added.

A similar effect was observed in ethanol at 200 ◦C when 30 mM2SO4 was replaced by 60 mM HCl. At equal normality the con-luding pH increased from 2.0 to 3.6 and the ethyl glucosidesield dropped by an order of a magnitude. An increased initial HCloncentration of 100 mM led to a concluding pH of 3.2 and thelcoholysis was improved only a little (Table 3).

. Conclusions

Lignocellulosic biomass can be converted in a single step viacid catalysed alcoholysis using small chain, linear alcohols yieldingainly furfural and alkyl glucosides. Maximum yields simul-

aneously obtained from wheat straw are 40% furfural (xylosequivalents) and 56% methyl glucosides (glucose equivalents). Theajor by-products are levulinic acid and its alkyl ester. The amount

f insoluble humins that are formed is limited.The depolymerisation and alkylation of carbohydrates (chains)

re mainly Brønsted acid catalysed and the presence of proton con-uming compounds, such as acid-neutralising native minerals orhloride, hamper the liquefaction efficiency and the alkyl gluco-ides yields. The optimum acid dose needs to be adjusted to thecid neutralisation capacity of the biomass.

A second important parameter is the process temperature. Inhis study, the best results were obtained around 175 ◦C. Theptimum temperature will depend on the heating and coolingransients of the equipment. The amount of dimethyl ether formedrom methanol under these conditions was limited, but becameeadily more prominent at higher temperatures. Solvent loss dueo ether formation from ethanol condensation was much smaller.he incubation time had only a limited effect.

Delignification prior to alcoholysis did not influence the resultsuch, but allowed higher product concentrations and enables a

eparate lignin valorisation.

cknowledgements

The authors acknowledge Arjan Smit and Ron van der Laan forheir contribution to this work and Ben van Egmond and Karinaogelpoel-de Wit for their analytical work. This work was fundedy Agentschap NL in the framework of the EOS-LT Catfur projectEOS-LT 08030). Additional funding from the Dutch Ministry of Eco-omic Affairs as part of the Biomass research programme of ECN iscknowledged.

eferences

[1] T. Werpy, G. Petersen, in: Top value-added chemicals from biomass, in: Vol. I– Results of Screening for Potential Candidates from Sugars and Synthesis Gas,US Department of Energy and Renewable Energy, Battelle, 2004.

[[

oday 223 (2014) 3– 10

[2] J.H. Holladay, J.F. White, J.J. Bozell, D. Johnson, Top value-added chemicalsfrom biomass, in: Vol. II – Results of Screening for Potential Candidates fromBiorefinery Lignin, US Department of Energy and Renewable Energy, Battelle,2007.

[3] B. Kamm, P.R. Gruber, M. Kamm, Biorefineries, Industrial Processes and Prod-ucts, Wiley-VCH, Weinheim, 2006.

[4] O. Casanova, S. Iborra, A. Corma, Journal of Catalysis 265 (2009) 109–116.[5] P. Walt, G.V. Vladimir, Advanced Synthesis & Catalysis 343 (2001)

102–111.[6] G.-J. Gruter, E. Jong de, Biofuels Technology 1 (2009) 11–17.[7] Y. Román-Leshkov, C.J. Barrett, Z.Y. Liu, J.A. Dumesic, Nature 447 (2007)

982–986.[8] M. Watanabe, Y. Aizawa, T. Iida, T.M. Aida, C. Levy, K. Sue, H. Inomata, Carbo-

hydrate Research 340 (2005) 1925–1930.[9] B. Girisuta, J.P.B.M. Janssen, H.J. Heeres, Chemical Engineering Research and

Design 84 (2006) 339–349.10] J. Zakzeski, R.J.H. Grisel, A.T. Smit, B.M. Weckhuysen, ChemSusChem 5 (2012)

430–437.11] J.B. Binder, R.T. Raines, Journal of the American Chemical Society 131 (2009)

1979–1985.12] R.-J. van Putten, J.C. van der Waal, E. de Jong, C.B. Rasrendra, H.J. Heeres, J.G. de

Vries, Chemical Reviews 113 (2013) 1499–1597.13] C. Fan, H. Guan, H. Zhang, J. Wang, S. Wang, X. Wang, Biomass and Bioenergy

35 (2011) 2659–2665.14] R. Karinen, K. Vilonen, M. Niemelä, ChemSusChem 4 (2011) 1002–1016.15] C. Rasrendra, I. Makertihartha, S. Adisasmito, H. Heeres, Topics in Catalysis 53

(2010) 1241–1247.16] S. Dutta, S. De, M. Alam, M.M. Abu-Omar, B. Saha, Journal of Catalysis 288 (2012)

8–15.17] V. Degirmenci, E.A. Pidko, P.C.M.M. Magusin, E.J.M. Hensen, ChemCatChem 3

(2011) 969–972.18] S. Lima, P. Neves, M.M. Antunes, M. Pillinger, N. Ignatyev, A.A. Valente, Applied

Catalysis A 363 (2009) 93–99.19] Y. Su, H.M. Brown, X. Huang, X. d. Zhou, J.E. Amonette, Z.C. Zhang, Applied

Catalysis 361 (2009) 117–122.20] G.J.M. Gruter, F. Dautzenberg, patent PCT WO 2007104514 (Sep 2007).21] M. Bicker, D. Kaiser, L. Ott, H. Vogel, Journal of Supercritical Fluids 36 (2005)

118–126.22] K. Garves, Journal of Wood Chemistry and Technology 8 (1988)

121–134.23] X. Hu, C. Lievens, A. Larcher, C.-Z. Li, Bioresource Technology 120 (2011)

10104–10113.24] X. Zhao, K. Cheng, D. Liu, Applied Biochemistry and Biotechnology 82 (2009)

815–827.25] W.J.J. Huijgen, A.T. Smit, J.H. Reith, H. Den Uil, Journal of Chemical Technology

and Biotechnology 86 (2011) 1428–1438.26] E. Muurinen, Organosolv Pulping. A Review and Distillation Study Related to

Peroxyacid Pulping, University of Oulo, Oulo, Finland, 2000, PhD thesis.27] M. Mascal, E.B. Nikitin, ChemSusChem 3 (2010) 1349–1351.28] J. Wildschut, A.T. Smit, J.H. Reith, W.J.J. Huijgen, Bioresource Technology 135

(2013) 58–66.29] National Renewable Energy Laboratory (NREL), Chemical analysis

and testing laboratory analytical procedures, Golden, CO, USA, 2009,http://www.nrel.gov/biomass/analytical procedures.html.

30] Technical Association of the Pulp and Paper Industry (TAPPI) Standard TestMethods. Norcross, GA, USA, 2009; http://www.tappi.org/2009

31] European Committee for Standardization. CEN/TS 14429: Characterizationof waste–leaching behaviour tests–influence of pH on leaching with initial

33] L. Paszner, H.J. Cho, TAPPI Journal 72 (1989) 135–142.34] Hawkes, Journal of Chemical Education 73 (1996) 516–517.